J. Kominek. Codebook reduction in fractal image compression. In Proceedings from IS&T/SPIE 1996 Symposium on Electronic Imaging: Science & Technology - Still Image Compression II, Volume 2669, 1996.
Through fractal compression is often characterized as a variant of vector quantization, the important step of codebook training has not migrated over. Unlike traditional VQ, the vector codebook is formed not prior to compression, but during. This often results in a number of similar, yet distinct, domain blocks. The result can be an unnecessarily large file. Coding efficiency is improved by pruning the codebook of redundant entries, with little or no resultant loss in image quality.
@InProceedings{Komi96,
Author = {Kominek, J.},
Title = {Codebook reduction in fractal image compression},
BookTitle = {Proceedings from IS&T/SPIE 1996 Symposium on Electronic Imaging: Science & Technology - Still Image Compression II},
Volume = {2669},
Month = {},
Year = {1996}
}